Patent Grant
6859063
The present invention relates to circuits and systems incorporating solid-state devices referred to herein as “transpinnors” and described in U.S. Pat. No. 5,929,636 and 6,031,273, the entire disclosures of which are incorporated herein by reference for all purposes. More specifically, the present application describes transpinnor-based transmitters and receivers for facilitating the reliable transmission of information via transmission lines and various applications thereof.
The vast majority of electronic circuits and systems manufactured and sold today are based on semiconductor technology developed over the last half century. Semiconductor processing techniques and techniques for manufacturing integrated circuits have become increasingly sophisticated resulting in ever smaller device size while increasing yield and reliability. However, the precision of such techniques appears to be approaching its limit, making it unlikely that systems manufactured according to such technique will be able to continue their historical adherence to Moore's Law which postulates a monotonic increase in available data processing power over time.
In addition, as the techniques for manufacturing semiconductor integrated circuits have increased in sophistication, so have they correspondingly increased in cost. For example, current state-of-the-art integrated circuits require a large number of processing steps to integrate semiconductor circuitry, metal layers, and embedded circuits, an issue which is exacerbated by the varied nature of the materials being integrated. And the demand for higher levels of complexity and integration continue to grow. The technical difficulties facing the semiconductor industry are well summarized by P. Packan in the Sep. 24, 1999, issue of Science magazine beginning at page 33, incorporated herein by reference in its entirety for all purposes.
Finally, there are some applications for which conventional semiconductor integrated circuit technology is simply not well suited. An example of such an application is spacecraft systems in which resistance to external radiation is extremely important. Electronic systems aboard spacecraft typically require elaborate shielding and safeguards to prevent loss of information and/or system failure due to exposure to any of the wide variety of forms of radiation commonly found outside earth's atmosphere. Not only are these measures costly in terms of dollars and weight, they are not always completely effective, an obvious drawback given the dangers of space travel.
In view of the foregoing, it is desirable to provide electronic systems which facilitate higher levels of integration, reduce manufacturing complexity, and provide a greater level of reliability in a wider variety of operating environments. As described in the aforementioned U.S. patents, such electronic systems are made possible by the advent of the all-metal, multi-purpose circuit element referred to as the “transpinnor.” It is therefore also desirable to facilitate the reliable communication of signals and data between the various components of such systems.
According to the present invention, electronic circuits and systems based on an all-metal solid-state device referred to herein as a “transpinnor” address the issues discussed above. More specifically, an embodiment of the present invention provides transmission line drivers and receivers implemented using transpinnor technology.
Thus, the invention provides a driver for a transmission line. The driver includes a network of thin-film elements exhibiting giant magnetoresistance, and an input conductor inductively coupled to at least one of the thin-film elements. The driver is operable to generate an output signal which is a function of a resistive imbalance among the thin-film elements and which is proportional to a power current in the network of thin-film elements. At least one dimension of each thin-film element is configured with reference to the characteristic impedance of the transmission line.
According to another embodiment, the invention provides a receiver for a transmission line. The receiver includes a network of thin-film elements exhibiting giant magnetoresistance, and an input conductor inductively coupled to at least one of the thin-film elements. The receiver is operable to generate an output signal which is a function of a resistive imbalance among the thin-film elements and which is proportional to a power current in the network of thin-film elements. The receiver further includes a termination impedance in series with the input conductor. The value of the termination impedance relates to the characteristic impedance of the transmission line.
A transceiver comprising the aforementioned driver and receiver is also described.
According to yet another embodiment, the invention provides a transceiver for a transmission line. The transceiver includes a driver comprising a first network of thin-film elements exhibiting giant magnetoresistance, and a first input conductor inductively coupled to at least one of the thin-film elements in the first network. At least one dimension of each thin-film element in the first network is configured with reference to the characteristic impedance of the transmission line. The transceiver also includes a receiver comprising a second network of thin-film elements exhibiting giant magnetoresistance, and a second input conductor inductively coupled to at least one of the thin-film elements in the second network. The receiver further includes a termination impedance in series with the second input conductor. The value of the termination impedance relates to the characteristic impedance of the transmission line.
A further understanding of the nature and advantages of the present invention may be realized by reference to the remaining portions of the specification and the drawings.
a shows a multilayer GMR film.
b shows a typical resistance curve for a GMR film such as the one shown in
a is a schematic diagram of a first transpinnor configuration.
b shows a plot of the output voltage of the transpinnor of
c and 2d show two alternative structures for the multilayer GMR film of
a is a schematic diagram of a second transpinnor configuration.
b shows a plot of the output voltage of the transpinnor of
a and 10b show transpinnors to operate as an AND gate and an OR gate, respectively.
a and 12b show a transpinnor configured as a switch according to a specific embodiment of the invention.
a and 13b show another transpinnor configured as a switch according to another specific embodiment of the invention.
a and 14b show a circuit symbol of a transpinnor switch and a circuit diagram of three transpinnor switches connected in series.
a and 16b show 2 transpinnor-based switching matrices.
Reference will now be made in detail to specific embodiments of the invention including the best modes contemplated by the inventors for carrying out the invention. Examples of these specific embodiments are illustrated in the accompanying drawings. While the invention is described in conjunction with these specific embodiments, it will be understood that it is not intended to limit the invention to the described embodiments. On the contrary, it is intended to cover alternatives, modifications, and equivalents as may be included within the spirit and scope of the invention as defined by the appended claims. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. The present invention may be practiced without some or all of these specific details. In addition, well known process operations have not been described in detail in order not to unnecessarily obscure the present invention.
“Giant magnetoresistance” (GMR) refers to the difference in the resistance that conduction electrons experience in passage through magnetic multilayer films which is dependent on the relative orientation of the magnetization in successive magnetic layers. For ferromagnetic materials, this difference occurs because the energy level for conducting electrons in a ferromagnetic layer is lower (by a few electron microvolts) for electrons with spin parallel to the magnetization rather than antiparallel. A GMR film is a composite structure comprising one or more multilayer periods, each period having at least two magnetic thin-film layers separated by a nonmagnetic conducting layer. A large change in resistance can occur in a GMR structure when the magnetizations in neighboring magnetic layers change between parallel and antiparallel alignments.
The property of giant magnetoresistance may be understood with reference to
GMR film 100 may be formed of one or more periods, each period having, for example, a cobalt layer characterized by a moderate coercivity, a copper layer, a permalloy layer characterized by a lower coercivity than the cobalt layer, and another copper layer. The different coercivities of the alternating magnetic layers make it possible to achieve an antiparallel orientation of the respective magnetization directions. The copper layers physically separate the magnetic layers, which otherwise would be tightly coupled by exchange forces. Consequently, it is possible to switch the magnetization in the low coercivity film without switching the magnetization in the high coercivity film.
a shows a schematic diagram of a transpinnor 200 in which a GMR multilayer thin-film strip 202 is disposed in a bridge configuration with three resistive elements 204. A conductor 206 is wound around GMR film 202 for supplying a magnetic field thereto. An input signal is applied at terminals 208 and 210. Output terminals 212 and 214 give the output voltage, as indicated by a voltmeter. This configuration allows the output voltage to be zero as well as positive and negative. It should be noted that the output of transpinnor 200 may be represented by an ammeter rather than a voltmeter. That is, the transpinnor output may be either a voltage, i.e., operation in an “open-circuit” mode, or a current, i.e., operation in a “short-circuit” mode.
As is readily apparent, the input (between terminals 208 and 210) is completely isolated electrically from the output (between nodes 212 and 214) even for a DC input current I. The magnitude of the output is proportional to the applied B+ voltage and is limited only by the current carrying capacity of GMR film 202.
As mentioned, the output voltage or current of transpinnor 200 changes as the resistance of GMR film 202 changes and is proportional to the voltage drop across GMR film 202 as the current passes through it. The output voltage or current can be bipolar or unipolar, depending on the ratios of resistances chosen for the other legs (i.e., the bias can be positive, negative, or zero). Also, depending on the squareness of the B-H loop, the output voltage or current can either be linear or a threshold step function. In addition, if the GMR film 202 is constructed symmetrically about the center, the net magnetic field from the current passing through the film will be zero. Therefore, the only limits on magnitude of the current are the heating of GMR film 202 and/or electromigration. The GMR films may employ metals having high electromigration thresholds, such as copper, cobalt, nickel and iron.
a shows a schematic diagram of another transpinnor 300 having a different configuration. Instead of only one GMR film, transpinnor 300 employs four GMR films 302 arranged in a bridge configuration with conductor 304 wound through them for supplying a magnetic field thereto. As with transpinnor 200, the input of the device (between terminals 306 and 308) is isolated electrically from the output (between nodes 310 and 312) even with a DC input current. Also, the output voltage of transpinnor 300 is determined by the magnitude of B+ and the current carrying capacity of GMR films 302. As shown in
The solid curve of
The GMR transpinnor of
It is desirable in particular applications for the GMR transpinnor to have a gain greater than unity. The low-frequency gain of GMR transpinnors is a function of their fundamental parameters. Referring again to
X=GMR/(100Hc) (1)
where Hc, represents the coercivity of the permalloy in the GMR film. The voltage gain of the GMR transpinnor of the present invention is proportional to the resistibility, and the power gain is proportional to the square of the resistibility.
The input line of the transpinnor produces a field. The ratio of field to the current by which it is produced is referred to herein as the coil efficiency, E. Generally speaking, the value of E increases dramatically as the size of the transpinnor decreases. If other parameters (including the resistance of the input line) stay the same, the voltage amplification is proportional to E, and the power amplification is proportional to the square of E.
Given the definitions of the various parameters of the transpinnor, the voltage amplification is given by
Avoltage=(R/r)IEX (2)
and the power amplification is given by
Apower=(R/r)I2E2X2 (3)
From (1) and (3) it becomes evident that the power amplification of transpinnor 500 is proportional to the square of the current, to the square of the GMR, to the square of the drive line efficiency, and inversely proportional to the square of the coercivity of the GMR film.
Some numerical examples of power amplification may be instructive. According to a first example, the input resistance is 0.8 Ohms, the resistance of the GMR film elements is 120 Ohms, the resistibility is 0.011/Oe, and the coil efficiency is 20 Oe/amp. If an input current of 500 mA is used, according to (3), the power amplification is 1.8. This is not a particularly good film.
According to a second example, the parameters are the same as for the first example above, except that the resistibility is 0.19/Oe. Now the power amplification is 541. This is higher than desirable for a logic tree, but may be reduced to a desirable value by appropriately decreasing the current.
According to a third example, a miniaturized transpinnor is configured as shown in
The conclusion is that substantial power amplification can be achieved with GMR transpinnors using existing GMR film configurations. Additionally, amplification factors in the hundreds can be obtained regardless of whether the transpinnors are large or so small as to be at the limits of conventional lithography because the power amplification factor is independent of the size of the device. However, although GMR transpinnors scale so their power amplification doesn't degrade when the devices are miniaturized, the power handling capability of the devices diminishes, of course, as the device size diminishes. GMR transpinnors can be designed to give either high output current and low output voltage, or high output voltage and low output current. These parameters are determined by the aspect ratio of the GMR film. If the GMR film is a long narrow conductor, the output is high voltage and low current. If the GMR film is a short wide conductor, the output is low voltage and high current. The power amplification is relatively independent of the aspect ratio.
To get high power amplification, the following may be done:
Although low GMR films with very low coercivity can be used to construct GMR transpinnors with high power amplification, the resulting device may be inefficient. If overall power consumption is a consideration, one should use high GMR films. It is possible, for example, to make GMR films with GMR of more than 22%.
There are a wide range of applications for which the transpinnor represents a significant advance. For example, transpinnors may be employed to implement nonvolatile logic gates, i.e., gates which maintain their states when power is removed. Additionally, because all-metal films exhibit much greater resistance to damage by radiation than semiconductors, transpinnors may be employed to implement intrinsically radiation-hard electronics.
The curve shown in
dR/dHL=(GMR)R/(HT−Hk) (4)
where GMR is the maximum resistance change, and HT must be larger than the maximum Hk of any region of the film. This differential resistance change can be quite large if the inhomogeneity of the film is small, and the corresponding amplification can be large. This is a sensitive method of achieving anhysteretic GMR films by a transverse-biased permeability. It results in an analog signal with a linear response within a certain range.
In another approach to eliminating the hysteresis, the permalloy layer in the transpinnor is driven and sensed in the hard direction. The cobalt layer is deposited so that its easy axis is parallel to the hard axis of the permalloy. this is accomplished by saturating the cobalt layer during its deposition at 90 degrees from the easy axis of the permalloy. This method does not generally require a bias field during operation; the exchange bias between the high coercivity layer(s) and the permalloy layer is normally sufficient to prevent the hard-axis loop from opening. The sensitivity of the hard-axis-driven film is not as good as in the approach based on the transverse-biased permeability (described above), but the linearity extends over a broader range and this method is easier to implement in that it avoids biasing in the hard direction and driving in the easy direction.
Yet another approach involves a sampling method. A pulse is applied to the transpinnor between each data sample. The pulse is of sufficient amplitude to saturate the permalloy layers in the transpinnor to an initial state that is the same regardless of whatever signal was applied in between. The frequency of the applied pulse should be higher than the highest frequency of interest in the signal to be amplified. The result of using narrow pulses to reinitialize the magnetic material before each data sample is to erase the magnetic history and to eliminate the hysteresis in the output. The output can be sensed either with sampling techniques or as an analog output with a low-pass filter.
It is generally understood that all possible electronic circuits, analog and digital, can be implemented using active components, e.g., transistors, in combination with four basic passive components, i.e., resistors, capacitors, inductors and transformers. It is also well known that neither inductors nor transformers are available in semiconductor bipolar technology. By contrast, the GMR transpinnors can be employed to provide both of these components. In fact, they are well suited to provide the basis of a variety of analog, digital and mixed general-purpose all-metal circuits, subsystems and systems. Since capacitance and resistance can be implemented with the same metal technology as that used for the passive transformer and the transpinnor, all these components can be combined very effectively on the same substrate to produce a comprehensive variety of all-metal circuits. Unlike semiconductor chips, whose performance suffers below a critical size, the characteristics of GMR devices improve as the dimensions are decreased.
Biased in the appropriate operating region, GMR transpinnors can be used as basic building blocks of logic gates, thereby providing the foundation for GMR-based digital electronics. While logic elements can be made with combinations of transpinnors, just as with transistors, there is another alternative. Various logic operations can be implemented with a single transpinnor. These transpinnors have more than one input line. Examples of such transpinnors are shown in
As mentioned above, when a transpinnor is balanced, its output is zero. An input current which exceeds the threshold for switching a lower-coercivity layer in one or more of the GMR films can change the film resistance, thus unbalancing the transpinnor, resulting in an output signal. Particular types of logic gates can be realized from the basic transpinnor by specific configurations of input lines and by suitable choices of input current values. Additional characteristics affecting the operation of transpinnor logic gates include the choice of resistors through which a given input current passes, the current polarities in selected resistors, and the direction of the magnetic field produced by the input current relative to the magnetization of the lower-coercivity layers in the transpinnor.
Two procedures are useful in implementing logic gates with a single transpinnor. One involves setting the transpinnor threshold which is determined by the coercivity of the low-coercivity layers in the GMR film. Various ways of establishing the coercivity of a thin film are known in the art. Thus, the threshold is set by choosing or adjusting the coercivity of at least one of the low-coercivity layers in the GMR films of the transpinnor. The other procedure involves switching the polarity of the GMR films which is determined by the magnetization orientation of all the film layers. The polarity of the transpinnor is thus switched by reversing the direction of magnetization of all layers of all GMR films in the transpinnor.
According to various embodiments, the balancing of transpinnor GMR elements is accomplished using a technique known as magnetoresistive trimming in which the magnetization of selected GMR elements are manipulated to achieve the desired balance. Magnetoresistive trimming techniques are described in International Publication No. WO 02/05470 A2 entitled MAGNETORESISTIVE TRIMMING OF GMR CIRCUITS published Jan. 17, 2002, the entire disclosure of which is incorporated herein by reference for all purposes.
Logic operations which can be implemented with a single transpinnor include the following:
AND gate: A transpinnor will not switch unless the sum of fields from the input lines exceeds the switching threshold. An AND gate is defined as one that yields no output unless all of its inputs are logical “1”s. If the transpinnor has n input lines, and the amplitude of each input pulse is (1/n)th of the threshold, then the transpinnor is an AND gate.
NAND gate: This is the inverse of the AND gate and gives an output if and only if all inputs are zero. A transpinnor NAND gate is made similarly as the AND gate, by reversing the magnetization of all elements so that the gate will just switch if all n inputs are logical “0”s and not switch if one or more are a logical “1”.
OR gate: The definition of an OR gate is one that gives an output if one or both inputs are a “1”. This can be made by setting the threshold of a transpinnor such that a single input is sufficient to switch the film.
A practical problem is presented by the fact that different switching thresholds are required for different single transpinnor logic gates. There are, however, a variety of ways in which these thresholds may be adjusted for different types of gates on the same substrate. These include manipulation of the order of deposition because the order strongly influences the coercivity of both the low and high coercivity films. This method involves additional deposition steps. Another method of adjusting the switching threshold for a particular transpinnor is derived from the fact that the magnetic field from a current carrying stripline depends on the width of the strip line.
NOR gate: The definition of a NOR gate is one that gives an output if one or both inputs are a “0”. This is merely the inverse of an OR. This can be done by reversing the polarity of the GMR films as in the above case of a NAND.
NOT gate: A NOT gate is an inverter that changes the polarity of an input pulse from positive to negative and vice versa. This is easily done with a transpinnor by reversing the polarity of the input winding, or by interchanging the power terminals.
Exclusive OR (XOR) gate: This is a gate that gives an output if one and only one of the inputs is a “1”. This can be done with a transpinnor such that one input is sufficient to switch the low-coercivity element, yielding an output, while two or more pulse inputs yield a field large enough to switch the high-coercivity element as well, yielding zero output. The gate must be reset after each use.
A circuit diagram of a transpinnor-based XOR gate 900 is shown in FIG. 9. As shown, input current 1 goes through resistors R1 and R3 and input current 2 goes through resistors R2 and R4. If the currents in both inputs are less than the switching threshold, the output is zero. If the current in one and only one of the two input currents is above this threshold, then the resistance of either pair of resistors changes, the transpinnor becomes unbalanced, and an output signal is generated. If both input currents are above the switching threshold, all four resistors change equally (if properly trimmed), the transpinnor remains balanced, and the output signal is zero.
A circuit diagrams for other transpinnor configurations are shown in
If the current in one, and only one, input is above the switching threshold, all four GMR elements change equally, the transpinnor remains balanced, and the output of gate 1000 is zero. If, on the other hand, the currents in both inputs are above the switching threshold (and thus the net current through R2 and R4 is below the switching threshold), the transpinnor becomes unbalanced and gate 1000 produces an output signal.
Referring now to
If currents in both inputs are less than half the switching current, all four GMR elements remain unchanged, the transpinnor remains balanced, and the output of OR gate 1050 is zero. However, if the current in either or both of the inputs are above the switching current, the resistances of elements R1 and R3 change while those of R2 and R4 remain the same, the transpinnor becomes unbalanced, and OR gate 1050 generates an output signal. It will be understood that the net current through R2 and R4 should not be sufficient to produce a magnetic field which could switch the lower-coercivity layers of these elements.
For digital applications, transpinnors with sharp thresholds and square-pulse outputs are desirable. For analog applications, a linear response is better. Transpinnors operating in the linear region can be used to develop a full complement of basic analog circuits, sufficient to create general-purpose analog circuitry based on GMR films.
A specific example of a transpinnor operating in the linear region for application to signal amplification illustrates some of the unique advantages of the dual functionality of the transpinnor over silicon technology. Differential amplifiers are typically used to eliminate common-mode signal and common-mode noise within the frequency range of their operation. As discussed above, the range of operation of the transpinnor in its transformer function extends from (and including) dc to the high-frequency cutoff limit. The GMR transpinnor can advantageously be utilized in its transformer function to remove common-mode signal in the differential-input mode, as well as in its transistor function to amplify a low signal in the single-ended output mode. In low-signal amplification, GMR transpinnors have the additional advantage of eliminating the problem of offset voltage at the input that is so troublesome in silicon integrated circuits. It should be noted that a high premium is paid in silicon technology to achieve low-offset input voltage for integrated differential amplifiers. That is, low-offset input voltage is achieved in silicon circuits only at the expense of degrading other parameters. No such price is associated with the use of transpinnors because of their dual transformer/transistor properties. Specifically, the input signal is applied to a differential input having the properties of a transformer primary with an additional advantage of flat low-frequency response inclusive to dc. The output signal is amplified by an output having transistor properties. Transpinnors are thus especially well suited as differential amplifiers.
Since transpinnors are current driven devices, an important parameter is the output current of a given transpinnor for a given input current. This determines whether one transpinnor can switch another, for example, or how much amplification can be achieved. Of particular interest is the dependence of the amplification factor A=iout/iin on the power supply to the transpinnor and on its parameters. This relationship is given by:
A=π1000gmr VL/(Hcw2Rsq) (5)
where V is the power supply voltage in volts, gmr is the fractional GMR value of the film (i.e., the GMR value is normally quoted as a percentage), Hc is the coercivity in Oe, w and L are the GMR strip width and length in microns, and Rsq=r/(L/w) is the sheet resistivity in ohms per square of a GMR film with resistance r (ohms per square is a standard term in thin film technology because the resistance from edge to edge of a thin film square is independent of the size of the square).
The field H produced by iin in a stripline of width w is given by:
H=2πiin/w (6)
and iout is given by:
iout=103gmr v/(2r) (7)
where H is in Oe, iin and iout are in mA, w is in microns, and V is in volts.
Many transpinnor-based devices require one transpinnor to switch another transpinnor. Examples include a transpinnor shift register, a transpinnor selection matrix, and a transpinnor multistage amplifier. When a transpinnor is used to switch another transpinnor, the output current of the switching transpinnor becomes the input current of the transpinnor to be switched. A single transpinnor can readily switch multiple transpinnors as shown by the following numerical examples of the performance characteristics of several transpinnor-based devices:
The foregoing examples illustrate that even transpinnors with modest GMR values can achieve enough gain to perform the analog and logic functions required to implement a wide variety of circuits including, for example, a field programmable gate array (FPGA) and a field programmable system-on-a-chip (FPSOC) as will be described below.
According to various embodiments of the invention, transpinnors may be configured to operate as switches that are nonvolatile like EEPROMs yet are characterized by the programming speed associated with SRAMs.
As described above and generally speaking, transpinnor technology operates by impressing a specific magnetization direction on a lower-coercivity layer (e.g., permalloy). In digital applications, two opposing magnetizations of this layer correspond to two logic levels. By contrast, the higher-coercivity layer (e.g., cobalt) in such applications remains pinned to a particular magnetization.
As will be discussed, reconfigurability of a programmable SOC may be achieved through a transpinnor switch which includes two nonmagnetic, conductor layers inductively coupled to various ones of the GMR films of which the transpinnor is composed, e.g., FIG. 12. In general and as discussed above, the current in the input conductor of the transpinnor switch is large enough to set (reverse) the magnetization of the lower-coercivity layers of the GMR films to which the input conductor is coupled but not that of the higher-coercivity layers. By contrast, the current in the switch conductor can be large enough to set (reverse) the magnetization of both layers in the GMR films to which the switch conductor is coupled.
The basic idea of the transpinnor switch in digital applications is to use the switch current to control the magnetization of the higher-coercivity layers in selected GMR films of the transpinnor in such a way that the output is either a logic signal following the input, or zero irrespective of the input. It should be noted that there are various ways of configuring the transpinnor switch to realize this functionality. That is, non-magnetic conductors may be inductively coupled to various subsets of the GMR films to achieve this functionality and remain within the scope of the invention.
According to one embodiment shown in
Transpinnor switch 1200 is initialized with all magnetizations in all four films parallel to one another (not shown). In this initialized state the resistances in all four films are low and the transpinnor is balanced, so there is no output when a power voltage is applied. A current is then applied via input conductor 1202 to reverse the magnetization in the low coercivity layers 1206 in films R1 and R3, resulting in a magnetization which is antiparallel to that of high coercivity layers 1208 as shown in
To turn switch 1200 “off,” a switch current is applied via switch conductor 1204 to set the magnetizations in the high and low coercivity layers of R1 and R4 in a direction opposite the previous magnetization state of the high coercivity layers. Then, a small current is applied via switch conductor 1204 to reverse the low coercivity (but not the high coercivity) magnetizations in R1 and R4. Finally, a current with the same polarity as that used to orient the low coercivity layers in R1 and R3 in
According to various embodiments, several transpinnor switches designed according to the invention may be connected in series to route a single input signal to a variety of circuits within a system. For example, a two-switch device may be configured with one switch “on” and the other “off” such that one of two circuits in a programmable system is enabled while another is disabled. As described above, the low coercivity layers in the “off” switch do not function, i.e., the output current is zero irrespective of which of the two logic levels is asserted at the input, while those in the “on” switch do, i.e., the output signal corresponds to the input signal. However, when the high coercivity layers of the appropriate films in both switches are magnetized in the other direction, the roles of the low coercivity layers in the two switches are reversed. Reconfiguration is thus achieved by simultaneously turning the “on” switch “off” and vice versa.
It should be noted that, according to some embodiments, the transition between balanced and unbalanced transpinnor configurations—and hence between output and no output—can be realized by reversing the polarity of the input signal to set the magnetization of the soft layer alone. However, this would obviate the use of a signal of given polarity to operate the logic circuitry in a given configuration.
According to various embodiment of the invention, there are a variety of ways in which transpinnor switch 1200 may be reconfigured. For example, we have just discussed shutting off the switch by reversing the magnetization of the high coercivity layers in R1 and R4. This shuts off the switch regardless of the input to the low coercivity layers of the films with which the input conductor is associated. It should be understood that one could just as well reverse the high coercivity layers in R2 and R3 to achieve the same effect.
If one instead reverses the high coercivity layers in R1 and R3, this merely reverses the polarity of the output, i.e., the switch conducts with full output for one input polarity. The same is true for switching the high coercivity layers of R2 and R4.
If one reverses the high coercivity layer in only one of the films, the result is a switch with half the output. Reversing the magnetization of the low coercivity layers with an input current does not shut the switch off. Reversing the high coercivity layers in three of the GMR elements (e.g., R1, R2, and R3) also reduces the output by half.
In summary, the two schemes that turn switch 1200 off for both input polarities are either to reverse the magnetization of the high coercivity layers in R1 and R4 or to reverse the magnetization of the high coercivity layers in R2 and R3.
Transpinnor switch 1200 has an input conductor 1202 coupled to only two thin-film elements, i.e., R1 and R3. Transpinnor switches designed according to various other embodiments of the invention may have an input conductor over all four thin-film elements in the bridge. Such a switch 1300 is shown in
Transpinnor switch 1300 is shown in
It should be noted that, according to various embodiments, the transpinnor switch of the present invention may be used for either digital or analog applications. According to an embodiment in which a transpinnor switch is employed to transmit analog signals, the switching field generated by the switching current is perpendicular to the easy-axis of the GMR elements rather than parallel. When the switch is enabled, the switching bias field is raised above the anisotropy field of the lower-coercivity layer. This causes the transpinnor to operate like a linear, nearly lossless transformer. When the bias field is turned off, there is no output unless the signal is large enough to exceed the coercivity of the lower-coercivity layer. In this way, analog signals may also be routed point-to-point according to the invention.
According to various embodiments of the invention, the transpinnor switch of the present invention may be configured to output an amplified version of the input. That is, using the techniques described above, transpinnor switches may be configured to provide a wide range of amplification factors including negative amplification factors, i.e., a transpinnor switch may be configured as an inverter. This amplification capability can be important for applications in which it is desirable to cascade transpinnor switches (see
a shows a circuit symbol representing a transpinnor switch 1400 having an input conductor 1402, a switch conductor 1404, and an output 1406.
The ability to cascade transpinnor switches is also advantageous for creating switching matrices such as the 2×2 switching matrix 1500 shown in FIG. 15. More specific implementations of some exemplary switching matrices will now be described with reference to
a shows a specific implementation of a 2×2 switching matrix 1600 corresponding to the simplified representation of FIG. 15. Each of the inputs can be switched into any of the outputs. Thus it is possible, for example, to connect input 1 to output 1, and input 2 to output 2, or, input 1 to output 2 and input 2 to output 1. One can have as many inputs to a transpinnor as desired, because an input line may be implemented as a stripline deposited above a GMR film and insulated therefrom, and one can have many such lines, one above another.
Programmable logic devices (PLDs) are a class of circuits widely used in LSI and VLSI design to implement two-level, sum-of-products Boolean functions. PLDs include programmable array logic (PALs), field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), and read only memories (ROMs). One of the advantages of PLDs is their highly regular layout structure. That is, a typical PLD includes an AND plane followed by an OR plane. The logic function performed by the device is determined by the presence or absence of contacts or connections at row and column intersections in a single conducting layer.
FPGAs are the most flexible of the PLDs in that they can be reconfigured multiple times. In conventional semiconductor technology, FPGA implementations typically choose between two very different types of memories to control their switches, one which ensures nonvolatility, a one which ensures speed. A block diagram of a generic FPGA architecture is shown in FIG. 17. FPGA 1700 has n inputs, k product terms (AND array 1702), m sum terms (OR array 1704), m output, and (2 nk+mk+2 m) switches. In conventional semiconductor technology, the switches are typically implemented with MOSFETs in conjunction with either EEPROMs or SRAM which may be reloaded with different data to reconfigure the FPGA. A significant drawback with the MOSFET/SRAM combination is that if power is lost, the SRAM must be reloaded for the FPGA to function properly. On the other hand, although EEPROMs avoid this issue because they are nonvolatile, they are an order of magnitude slower than SRAM, limiting the reprogramming speed of an EEPROM-based FPGA accordingly.
Therefore, according to the present invention, an FPGA architecture is provided which employs any of the various embodiments of the transpinnor switches described herein (e.g., transpinnor switch 1200) as the basis for the switch matrices with which the product and sum terms of the FPGA may be interconnected. The nonvolatile nature of the state of these transpinnor switches, and the speed with which they may be accessed and switched results in a solution which has the best of both previous options without the attendant disadvantages. According to some of these embodiments, the product terms and sum terms of the FPGA (e.g., arrays 1702 and 1704) are implemented using transpinnor logic gates including, for example, those described above with reference to
According to various embodiments, an FPGA architecture designed according to the invention may conform to the conventional paradigm in which switches are controlled by associated memory elements. According to some of these embodiments, the memory elements may be implemented using any of the GMR-based memory cells described in U.S. Pat. No. 5,587,943 for NONVOLATILE MAGNETORESISTIVE MEMORY WITH FULLY CLOSED FLUX OPERATION issued Dec. 24, 1996, and International Publication No. WO 02/05268 A2 entitled ALL METAL GIANT MAGNETORESISTIVE MEMORY published Jan. 17, 2002, the entire disclosures of both of which are incorporated herein by reference for all purposes. Such memory elements will be referred to herein generically as SpinRAM elements.
According to other embodiments, the nonvolatile nature of the transpinnor switch of the present invention obviates the need for controlling the switches with associated memory elements. That is, because the state of the switches designed according to the invention is nonvolatile, the switches themselves may be directly programmed as opposed to indirectly programming them via the associated memory elements.
According to still other embodiments, one or more FPGAs designed according to the present invention is included as part of a larger, field programmable system-on-a-chip (FPSOC). One such generalized embodiment is shown in FIG. 18. According to various embodiments, any or all of the system components of FPSOC 1800 may be based on all-metal GMR electronics. For example, FPGA 1802 may be implemented as described above using transpinnor switch matrices and transpinnor logic gates. Some or all of the mixed signal components of field programmable analog array 1804 (e.g., differential amplifiers, sample-and-hold circuits, etc.) may be implemented using transpinnor-based circuits. In addition, memory array 1806 may be implemented as a SpinRAM array. Arithmetic logic unit 1808, multiply/accumulate unit 1810, and general purpose I/O 1812 may all be implemented using transpinnor logic gates.
A common method of reconfiguring a FPSOC in conventional semiconductor technology employs logic gates for routing configuration signals. According to one embodiment of the invention, such a method can also be implemented using the transpinnor logic gates described herein. According to other embodiments, advantage is taken of the unique aspects of transpinnor technology to provide a simpler approach using transpinnor switches to reconfigure such FPSOCs.
Generally speaking, PLDs, FPGAs, and FPSOCs designed according to the invention may implement any of the wide variety of functions and be employed in any of the wide variety of applications and environments as any of their conventional counterparts. In addition, for embodiments in which all of the circuit components, functional blocks, and subsystems are based on the all-metal GMR technology described herein, several advantages over conventional semiconductor or hybrid implementations will be enjoyed. That is, such all-metal circuits and systems are intrinsically radiation-hard. From a manufacturing standpoint, fewer processing steps, lower processing temperatures, and fewer masks, make such all-metal implementations logistically and economically superior. Single transpinnor implementations of conventionally more complicated circuits, e.g., logic gates, differential amplifiers, sample-and-hold circuits, comparators, etc., and the closed-flux nature of some memory elements facilitate increased density as well more reliable performance.
In some implementations of systems incorporating all-metal electronics, transmission line effects need to be taken into account. That is, for example, in systems in which signals are transmitted at high speeds or over long distances, it is important that the terminating impedances at the ends of a transmission line match its characteristic impedance in order to transmit the signals reliably and efficiently. This is particularly true, for example, in the complex architectures that can be implemented using GMR technology. More specifically, the transmission of information between multiple GMR chips in such systems will often need to deal with such transmission line effects.
A simple model for the transmission of signals over a line is given by its characteristic impedance, i.e., the instantaneous equivalent resistance that the transmitter sees at the source end when a signal voltage is applied, normally denoted Z0. It is a function of the separation distance, separating dielectric material, and geometry of the two conductors. In a uniform transmission line the characteristic impedance looking back from any point along the line is the same. A coaxial cable typically has the most uniform geometry and impedance, but even a twisted pair is sufficiently uniform to be very effective as a transmission medium. Typical characteristic impedances for lines used to transmit digital information are 50 to 200 ohms.
When both the source and destination are terminated with resistors that have the same value as the characteristic impedance, optimum power is transmitted and there are no signal reflections. Having no reflections is important to prevent successive signals from interfering at a destination node when the travel time over the transmission line gets to be greater than the cycle time of the signals. For example, if a signal cycle time is 1 ns (about 1 foot of travel) and the transmission line is more than 10 ft there can be as many as 10 signals traveling on the transmission line between the transmitter and receiver. Any excessive reflections will interfere with the reception of the real signals.
When a signal waveform is applied at the source node of a transmission line, it takes a predictable amount of time to travel down the line, i.e., the delay due to an electromagnetic wave traveling at the speed of light in the dielectric medium separating the two transmission conductors. A reasonable working approximation is to take the speed to be 1 foot/ns. The actual delay depends on the material between the conductors (e.g., the speed of light in conventional coaxial cable or twisted pair conductors is roughly 0.6 c, or about 1.5 ft/ns). For normal circuitry this is not an issue. The fact that the delays can be predicted allows interconnects to be designed with predictable delays and compensating circuits if the delays are significant. For short connections, transmission-line delays may be ignored as long as design rules governing these delays are in place. When multiple transmission lines carry parallel signals that need to be synchronized, skew in the arrival of the signals at their destination may have to be taken into account. Many solutions are available (including the lengthening of the shorter transmission lines) to ensure that there is minimal skew. When the clock speed of digital logic gets close to the transmission speed of signals between the logic components, interconnects may have to be treated as transmission lines even though they may only appear to be short printed-circuit traces or internal interconnects on a chip.
Reflections of signals are the main source of problems in short-distance transmission-line systems. When a transmission line is properly terminated at both ends with the same resistance as the characteristic impedance, a waveform (in this case a pulse) is repeated, with the proper delay, at any point in the transmission line. There is no reflection of the waveform. Two limiting cases of the terminating node are of interest, open and shorted. The waveform for a circuit open at the termination and properly terminated at the source is shown in FIG. 19A.
Until the round-trip delay is completed, the source sees the equivalent of a terminated transmission line. Since the energy arriving at the termination is not absorbed, it is reflected as a negative current that will cause the voltage at the source to overshoot. As successive reflections appear at the source, the waveform gradually dampens out to its steady state and the current flowing in the transmission line goes to 0. If the transmission line is very short compared to the signal length, it will become equivalent to a capacitor. When a fast pulse is applied, the initial response is due to the capacitor charging and, when charged, the transients die down. Therefore a short unterminated transmission line can be replaced with an equivalent capacitance for purposes of both modeling and design.
The second limiting case of interest is when the terminating node of a transmission line is shorted. The waveform for a pulse applied at the source node is shown in FIG. 19B. Each step represents a round-trip delay in the transmission line. Initially the line appears to the source as a normal terminated line and the normal current is sent. When the current arrives at the destination, it is not absorbed and is therefore reflected negatively. When arriving back at the source, the equivalent voltage is reduced. The bouncing back and forth continues until there is no voltage across the source terminals and the maximum current flows through both conductors of the transmission line.
The use of semiconductors as drivers and receivers for transmission lines has been developed over many years, and many commercial products are available. For example, standards-based system interconnects such as the RS485 are used heavily for interconnecting systems. There are families of RS485 transceivers that make implementing an interface to an RS485 bus very easy and commonplace. The advantage of such standards-based transmission-line systems is that they are vendor and technology independent.
More and more interconnects are gravitating toward serial transmission of information as a replacement for parallel lines at all levels of systems. The Universal Serial Bus (USB) is an example. The transmission line is only point-to-point, and the terminations are well matched to get very good transmission performance. Internal buses like the SCSI bus also depend on transmission-line characteristics of interconnects and well-matched terminations. In this case the drivers and receivers are built into the host, and peripheral subsystems are based on well-defined standards.
In SERDES (serializer/deserializer) schemes, an array of parallel inputs is sampled by a serializer and then sent at a very high speed to the other end of the transmission line, where a deserializer builds the parallel version and gates it onto the output as parallel signals. Many of these SERDES operate at a gigahertz, some are now running above 3 GHz, with plans to reach 10 GHz. The main advantage of this approach is that many parallel interconnects can be eliminated with a single transmission-line replacement. In systems incorporating all-metal electronics such as those described above, similar techniques may be employed.
As will become clear, all-metal solutions corresponding to such transceiver types (as well as many others) may be provided according to various embodiments of the invention. More specifically and according to such various embodiments of the invention, appropriately configured transpinnors are provided as drivers and receivers for transmission lines.
According to various embodiments, a variety of transpinnor characteristics may contribute to their suitability for implementation as transceiver components. For example, as described above, the output terminals of transpinnors are electrically isolated from their inputs to the extent of the strength of the dielectric between them. This helps inhibit unintended subsystem interactions such as ground loops and noise. In addition, the outputs of a transpinnor are the bridge terminals of a four GMR element bridge structure. As a result, as the transpinnor is switched, the resistance seen at the output terminals remains substantially constant. This assumes that the GMR elements are well balanced; a situation that should be easy to achieve since all four elements are adjacent and should have the same GMR response curve.
As discussed above, to prevent reflections and noise the source impedance of a transmission line driver should be the same as the characteristic impedance of the transmission line. Transpinnors can be configured by adjusting their length and width so that the output terminals present any desired impedance to the transmission line. Moreover, when the transmitting transpinnor switches polarity the output side power supply and ground terminals of the transpinnor are essentially noise-free. The reason for this is that the switching of the individual GMR elements in the transpinnor is done symmetrically and does not result in a significant impedance change from the perspective of the power supply. As result, little or no ground current perturbations are experienced by the power supply. This is in contrast to CMOS line drivers where rail-to-rail switching may induce large transient spikes producing power supply noise problems for the rest of the circuit.
When a transpinnor is configured as a receiver according to specific embodiments of the invention, its input is nearly equivalent to a short circuit. This allows multiple receivers to be deployed on the same transmission line thereby supporting large fan-outs. Transpinnor receivers can also be used as taps on a transmission line since they absorb little or no energy and therefore do not greatly affect the transmission line characteristic. It will be understood, that the low impedance input of transpinnor receivers means that the transmission line should be terminated by an impedance corresponding, for example, to the transmission line characteristic impedance.
Transpinnors may also be configured to have noise-tolerant receiver input characteristics. That is, because of the nature of these GMR devices, they can be configured as receivers such that small signals at the transpinnor input (e.g., from noise or transmission line reflections) can be tolerated by properly biasing and conditioning the GMR elements such that a strong enough signal must be received for the input to switch to the opposite polarity.
Another advantage of transpinnor-based transceiver components is that the GMR elements for both transmitters and receivers can be designed so that they have memory. In the case of receivers, this capability makes it possible to receive and remember very short transmitted pulses. This can be important, for example, in transmission line systems with very low power, or in control systems in which the amount of information to be transmitted is low and relatively infrequent but where the transmission power must nevertheless be minimized.
On the transmitter side, this allows the transmitter to be set by input pulses (e.g., the driving logic is pulse based), and allows the output side to be pulsed whenever convenient. This could be useful, for example, where it is desirable that the output pulse which produces the transmitted signal is independent of the timing of the corresponding input pulse, i.e., the transmitter acts as a holding register.
For transpinnor-circuit to transpinnor-circuit transmissions a variety of contexts will be encountered, e.g., traces connecting driving transpinnors with receiving transpinnors on a single chip, signal lines between chips routed via printed circuit boards (PCBs), and signal lines between PCBs or subsystems. The transmitting medium can be any of a wide variety of mediums including, for example, coaxial cable, twisted pair, or PCB stripline conductors.
According to a particular embodiment and as described above, a transpinnor driver is configured as a set of four GMR elements interconnected in a Wheatstone bridge. The GMR elements can be adjusted at design time (or even subsequently configured) such that the driver presents a source impedance to a transmission line which is at or near the characteristic impedance of the line, i.e., the transpinnor can be used to directly drive the transmission line. A circuit diagram for an exemplary transpinnor driver is shown in FIG. 20.
Each of the GMR elements in transpinnor driver 2000 has an impedance value Z0*(1−gmr/2), where Z0 is the characteristic impedance of the transmission line, and gmr is the fractional GMR value of the GMR elements. Driver 2000 may operate in a variety of manners depending on the particular application. According to one implementation, power is always applied to driver 2000 at terminals CP and CN, and the output signal is controlled by the input signal at terminals INP, INN. With a logic 0 input, e.g., 0 volts, driver 2000 is configured so that there is no voltage at the output terminals OUTP and OUTN. When the input is changed to a logic 1, e.g., 3-5 volts, the magnetizations of the GMR elements are flipped, the bridge becomes unbalanced, and a positive voltage appears at the OUTP, OUTN terminals. Thus, information is transmitted as positive pulses on the transmission line. It will be understood that the design can also be altered slightly to allow negative pulses to represent information.
According to another implementation, power is again always applied to transpinnor driver 2000 and either a negative or a positive output current flows depending on the state of the input. This mode allows the receiving nodes to detect if the driver is operational, i.e., the absence of current can indicate a special condition or a circuit fault.
According to yet another implementation, the power to the transpinnor at terminals CP and CN is only applied when a pulse is sent out. The input signal or the state of the GMR elements determines whether the transmitted pulse is negative or positive. This implementation is characterized by some significant advantages. By pulsing the output, energy is conserved. In addition, the fact that the output pulse is going either negative or positive both allows the information to be transmitted and allows the signal to be self-clocking. This approach does require a receiver which is able to capture and discriminate the incoming pulse. Therefore, according to another specific embodiment of the invention, a transpinnor receiver is provided which has such a capability.
Because the input of a particular implementation of a transpinnor is a stripline with essentially zero resistance, such transpinnors are natural transmission line receivers. That is, the input of such a transpinnor can be used as a tap on a transmission line with minimal theoretical interruption to the signal on the line.
According to various embodiments, the combination of the signal current in the transmission line and the sensitivity of the transpinnor input are sufficient to cause the soft layer of the GMR elements of the receiving transpinnor to switch. When this occurs, the output senses the change directly. According to specific embodiments, the receiving transpinnor can also be designed to have memory to store the input pulse and polarity. According to such embodiments, the transpinnor can be powered at a later time to read the state that was captured.
According to various embodiments of the present invention, both static and pulsed transceiver circuits may be constructed using transpinnor-based transmitters and receivers such as those illustrated in
In addition, the transmitter is designed to provide enough current drive to be able to switch the receiving transpinnors. For example, if Imin is the minimum receiving current, the driving voltage at the transmitter side should be at least Z0*Imin volts. Since this is the voltage produced at the transmitter bridge terminals, the actual driving voltage must be at least (Z0*Imin)*2/gmr where gmr is the fractional representation of the GMR value. For example, if the receiver switches fully with 2 ma of current, the driving voltage at the bridge terminals must be (50*0.002)=0.1 volts. However, the terminal equivalent driving voltage is 0.2 volts since both the transpinnor equivalent resistance and the transmission line equivalent resistances divide the voltage. For a transpinnor with a GMR of 5%, gmr=0.05, and the actual transmitter driving voltage for this example is then (0.2/gmr)=4.0 volts. Note that the better the GMR, the lower the driving voltage required. Note also that the transmission power efficiency is proportional to the GMR of the transpinnor.
A specific implementation of a static transpinnor receiver designed according to the invention can be characterized in the following manner. At a specified threshold current level such a transpinnor receiver switches output states. This threshold discriminates between logic levels on the transmission lines.
According to various embodiments of the invention, a pulsed transpinnor transceiver may be constructed with GMR elements that have memory. According to one such embodiment, the transpinnor elements of such a transceiver are designed with the characteristic curve of FIG. 23. This curve applies to both the transpinnor used as a transmitter and also as a receiver. According to a specific embodiment, the characteristics of the receiver are more carefully adjusted so that the switching threshold is as crisp as possible and the noise tolerance is at an acceptable level.
Where transpinnor 2400 is employed as a transmitter, VDP and VDN are pulsed to send a signal onto the transmission line. Where transpinnor 2400 is employed as a receiver, these two lines are activated to read the state of the transpinnor. Signal lines RXP and RXN are the output terminals. For the transmitter, they are connected to the transmission line and for the receiver, they are connected to other logic circuitry (which may comprise transpinnor-based circuitry) that processes the input data. Signal lines HCLRP and HCLRN are additional input controls which may be provided for controlling the proper biasing of the hard (Cobalt) GMR layer.
The TXR13 trace represents the state of the resistance in GMR elements 1 and 3. After a 1 is entered the resistance is at the high level corresponding to the high state. The resistances for GMR elements 2 and 4 are opposite polarities. The TXCLR trace represents the transmitter clear pulse. This pulse is sent after each bit is transmitted. It puts the transmitter into low state.
The TXP trace is the transmit pulse applied across the VDP and VDN transpinnor terminals. This pulse activates the output of the transmitter and produces either a positive or negative pulse on the transmission line depending on the state of the transmitter. Note that the output level versus input level is proportional to the GMR of the device. The outgoing pulse must be strong enough and long enough to switch the receiver transpinnors.
The TXLINE trace is the voltage and current waveform seen on the transmission line itself. It is typically delayed and degraded from the input pulse due to capacitances and irregularities in the transmission line. As the pulse travels down the transmission line, it must be strong enough to switch the receiving transpinnors. Note that a positive pulse is sent with a data bit of 1 and a negative pulse with a data bit of 0. It should be noted that this can be used in a more refined system to make the transmissions self-clocking.
On the receiver side, a clear pulse (the RXCLR trace) is used in between data bits to clear the receiver to a 0 state. This signal can be derived form the transmission line directly. Alternatively, it can be derived from a separate clock source that synchronizes transmits and receives. The RXR13 trace represents the state of the GMR resistance in the receiving transpinnor's GMR elements 1 and 3. Note that after a receipt of a positive pulse, the resistance becomes high and after a negative pulse, it remains low. The RXD trace indicates that receiver data are available. That is, this signal shows the periods in which the receiving transpinnor can be sampled for the received data.
For a full GMR implemented solution, all of the logic for buffering and sequencing of the transmit and receive portions of the system may also be implemented as GMR elements.
While the invention has been particularly shown and described with reference to specific embodiments thereof, it will be understood by those skilled in the art that changes in the form and details of the disclosed embodiments may be made without departing from the spirit or scope of the invention. For example, embodiments have been described above with reference to conventional interface types which the transceiver components of the present invention may be configured to emulate. It should be noted that any references to such interfaces are merely exemplary and should not in any way be construed to limit the scope of the invention.
It should also be noted, for example, that the basic transpinnor topology of the present invention is not limited to any particular magnetic materials, bridge configuration, number of thin-film periods, etc. That is, any materials, configurations, periodicity, etc., of thin-film structures which are appropriate to enable the functions described herein are contemplated to be within the scope of the invention. It is also important to note that the transceiver component configurations described herein are merely exemplary and that many functionally equivalent configurations are within the scope of the invention.
Finally, although various advantages, aspects, and objects of the present invention have been discussed herein with reference to various embodiments, it will be understood that the scope of the invention should not be limited by reference to such advantages, aspects, and objects. Rather, the scope of the invention should be determined with reference to the appended claims.
The present application claims priority from U.S. Provisional Patent Application No. 60/372,712 for A TRANSPINNOR-BASED TRANSMISSION LINE TRANSCEIVER AND APPLICATIONS filed on Apr. 11, 2002, the entire disclosure of which is incorporated herein by reference for all purposes.
| Number | Name | Date | Kind |
|---|---|---|---|
| 3972786 | Ballard | Aug 1976 | A |
| 4751677 | Daughton et al. | Jun 1988 | A |
| 5173873 | Wu et al. | Dec 1992 | A |
| 5561368 | Dovek et al. | Oct 1996 | A |
| 5565236 | Gambino et al. | Oct 1996 | A |
| 5585986 | Parkin | Dec 1996 | A |
| 5587943 | Torok et al. | Dec 1996 | A |
| 5640754 | Lazzari et al. | Jun 1997 | A |
| 5650889 | Yamamoto et al. | Jul 1997 | A |
| 5650958 | Gallagher et al. | Jul 1997 | A |
| 5652445 | Johnson | Jul 1997 | A |
| 5654566 | Johnson | Aug 1997 | A |
| 5686837 | Coehoorn et al. | Nov 1997 | A |
| 5929636 | Torok et al. | Jul 1999 | A |
| 6031273 | Torok et al. | Feb 2000 | A |
| 6169292 | Yamazaki et al. | Jan 2001 | B1 |
| 20020029462 | Spitzer et al. | Mar 2002 | A1 |
| Number | Date | Country |
|---|---|---|
| WO 9741601 | Nov 1997 | WO |
| WO 0205268 | Jan 2002 | WO |
| WO 0205470 | Jan 2002 | WO |
| WO 02078100 | Oct 2002 | WO |
| WO 03088283 | Oct 2003 | WO |
| Number | Date | Country | |
|---|---|---|---|
| 20030231077 A1 | Dec 2003 | US |
| Number | Date | Country | |
|---|---|---|---|
| 60372712 | Apr 2002 | US |
